Field of the Invention
The invention concerns a method for image acquisition with a magnetic resonance device using a magnetic resonance sequence (in particular a PETRA sequence) of the type wherein k-space corresponding to the imaging area is scanned, with a first region of k-space that does not include the center of k-space being scanned radially along a number of spokes emanating from the center of k-space, and wherein at least two phase coding gradients have already been ramped up completely before administration of the excitation pulse; and wherein a second central region of k-space that remains without the first region is scanned in a Cartesian manner (in particular via single point imaging), with a pre-pulse (in particular an inversion pulse to establish a T1 contrast) being radiated before a predetermined number of individual measurements for the purpose of increasing the contrast. The invention also concerns a magnetic resonance apparatus that implements such a method.
Description of the Prior Art
Sequences with ultrashort echo times (thus echo times TE<0.5 ms) offer new fields of application in magnetic resonance imaging. They enable the depiction of substances that are not visible with conventional magnetic resonance sequences (for example spin echo or gradient echo sequences) since their repetition time T2 is markedly shorter than the echo time and their signal has already decayed at the acquisition point in time. Some magnetic resonance sequences with ultrashort echo times are additionally extremely quiet since only extremely small gradient changes are necessary. Examples of such sequences that markedly reduce the noise exposure of the patient are the zTE (zero TE sequence), the WASPI sequence (Water and Fat Suppressed Proton Projection MRI), the SWIFT sequence (Sweep Imaging with Fourier Transformation) and the PETRA sequence (Pointwise Encoding Time reduction with Radial Acquisition).
A number of magnetic resonance sequences with ultrashort echo time have been proposed, for example the radial UTE sequence (“Ultrashort Echo Time”, see for example the article by Sonia Nielles-Vallespin, “3D radial projection technique with ultrashort echo times for sodium MRI: clinical applications in human brain and skeletal muscle”, Magn. Reson. Med. 2007; 57; Pages 74-81). After a wait time after an excitation pulse, the gradients are thereby ramped up and begun simultaneously with the data acquisition. The k-space trajectory that is scanned in such a manner after an excitation travels radially outward from the center of k-space. Therefore, before the reconstruction of the image data (by means of Fourier transformation) starting from raw data acquired in k-space, these raw data are initially transformed onto a Cartesian k-space grid (for example via regridding).
A further approach in order to enable short echo times is to scan (enter raw data into) k-space at points by detection of the free induction decay (FID) is detected. Such a method is also designated as single point imaging, since essentially only one raw data point in k-space is detected per radio-frequency pulse. An example of such a method for single point imaging is the RASP method (“Rapid Signal Point Imaging”, O. Heid. et al, SMR, 3rd Annual Meeting, Page 684, 1995). A raw data point in k-space is thereby read out at the echo time TE at a fixed point in time after the radio-frequency excitation pulse, the phase of which raw data point was coded by gradients which are changed by means of the magnetic resonance device for each raw data point or, respectively, measurement point so that k-space can be scanned point-by-point.
A further shortening of the echo time and of the total acquisition time is enabled by the PETRA sequence, which is described by DE 10 2010 041 446 A1 (corresponding to U.S. Pat. No. 8,887,533) and an article by D. Grodzki et al., “Ultrashort Echo Time Imaging Using Pointwise Encoding Time Reduction With Radial Acquisition (PETRA)”, Magnetic Resonance in Medicine 67, Pages 510-518, 2012. These publications are herewith incorporated by reference into the disclosure content of the invention, which concerns the concrete realization of PETRA sequences. In the PETRA sequence, k-space corresponding to the imaging area is read out in two different ways. A first region, which does not include the center of k-space, is scanned by at least two phase coding gradients being initially switched (activated) in respective spatial directions by a gradient system of the magnetic resonance apparatus. Only after reaching the full strength of the switched phase coding gradients is a non-selective radio-frequency excitation pulse radiated by a radio-frequency transmission/reception device of the magnetic resonance apparatus. After a time t1 after the last radiated excitation pulse, echo signals are acquired by means of the radio-frequency transmission/reception device (or an additional, possibly dedicated radio-frequency reception device) and these are stored as raw data points along the radial k-space trajectories (spokes) predetermined by the strength of the phase coding gradients. These steps are repeated until k-space corresponding to the imaging area is read out along radial k-space trajectories in the first region depending on time t1. The switching of the phase coding gradients and the wait until these are ramped up can be further reduced to the echo time, for example in comparison to the UTE sequence. However, a central, spherical region including the center of k-space—the second region of k-space—cannot be scanned in that the phase coding gradients have already been ramped up. It is consequently provided that the second region of k-space (which consequently is not covered in the first region of k-space and which includes the center of k-space) is measured differently, wherein the scanning here takes place in a Cartesian manner, in particular by means of a single point imaging method (for example RASP). Since the raw data acquired in this second portion of the scanning are already situated on a Cartesian k-space grid, while the radially read-out raw data must still be transformed into this (as was already presented above) before image data can be reconstructed from the raw data by means of Fourier transformation, an additional savings of cost and time results.
The contrast of magnetic resonance sequences with ultrashort echo time (in particular thus also the PETRA sequence) is in the range of proton density weighting to T1 weighting. Given constant repetition time and constant flip angle over the measurement, what is known as a steady state develops that determines the precise contrast. In the zTE, WASPI, SWIFT and PETRA sequence, the flip angles are often limited to less than approximately eight to twelve degrees, which leads to a predominantly proton density-weighted contrast given typical repetition times of 3 to 5 ms.
In order to obtain a T1 or also a T2 contrast, it has been proposed to use pre-pulses that are respectively radiated before at least one part of the measurement processes. To save time, it is thus conceivable to apply the pre-pulses only every n repetitions, which (for example) is described in the article “Quiet T1- and T2-weighted brain imaging using SWIFT”, Proc. ISMRM 2011, Page 2723 by R. Chamberlain et al.
For the MPRAGE sequence (see for example the article by M. Brant-Zawadzki et al., “MP RAGE: a three-dimensional T1-weighted, gradient-echo sequence—initial experience in the brain”, Radiology 182, Pages 769-775, 1992), individual k-space lines are scanned in a Cartesian manner. If pre-pulses are also used here, after the pre-pulse a defined time TVP is initially waited here, whereupon an acquisition duration of TACQ follows in which a number of n=TACQ/TR repetitions are measured, wherein TR designates the repetition time (as is typical). After the acquisition duration, a wait time can further be provided before the next pre-pulse is applied. During the wait time, the spins relax, which can possibly be advantageous for the signal-to-noise ratio, but a complete relaxation typically no longer occurs.
This is explained in detail using the example of an inversion pulse for the T1 weighting. In this case, the spins are initially inverted (consequently rotated by a flip angle of 180°) by the pre-pulse formed as an inversion pulse. If excitation pulses that concern a smaller flip angle are now provided in the relaxation (always spaced by the repetition time), a stability magnetization that does not correspond to the maximum transverse magnetization results depending on the relaxation of the respective material, given which stability magnetization the relaxation time is ultimately “stopped” by the excitation pulses, wherein this stability magnetization is different for different materials (for example grey and white brain matter). A T1 weighting results from this.
If the data acquisition is then interrupted for the new pre-pulses, a complete relaxation also does not occur, such that consequently a rotation out of the maximum transverse magnetization does not occur, but rather either a rotation directly out of the stability magnetization or by a value between the maximum transverse magnetization and the stability magnetization. A steady state therefore results after a specific time (a transient event), which means that the curves of the magnetizations are the same given each cycle of pre-pulse and measurement process.
In the MPRAGE sequence measurement (data acquisition) takes place only in the steady state, which has arisen at the beginning of the complete measurement after a few of these cycles (in part already after one cycle). Often a waiting takes place for the duration of provided a pair of these cycles before executing the measurement, in order to not contaminate the measurement with data from the transient event.
A procedure is known to optimize the MPRAGE sequence so that an optimally good contrast—for example between grey and white brain matter—is achieved given an optimally high SNR. In this procedure, an optimized point in time TI after the administration of the pre-pulse is determined, in which an optimally good contrast is provided (for example a clear difference between the transverse magnetization components) but also in which the absolute value (of the transverse magnetization components, for example) is large enough that the signal-to-noise ratio is sufficiently high. A balancing ultimately takes place, from the result of which an optimal point in time TI can be derived that typically is during the relaxation process, before reaching the stability magnetization discussed above.
For the MPRAGE sequence it was then proposed that the k-space lines that are closest to the k-space center, and those that are most decisive for the contrast and the signal-to-noise ratio, be specifically measured at the optimized point in time T1 after the administration of the pre-pulse.
Due to the different principle, this procedure in the MPRAGE sequence cannot be directly transferred to the PETRA sequence.